A Metaheuristic Approach for Image Segmentation Using Genetic Algorithm
Joy Bhattacharjee, Soumen Santra, Arpan Deyasi · Lecture notes in networks and systems · 2021
Euclidian distance between any two observations in M × N matrix is computed using genetic algorithm. RGB image is classified using K-means clustering, and Euclidian distance matrix is calculated involving fitness function. Higher limit for terminating the program are assigned for both normal (500) as well as for higher accuracy values (2000). Clusters are assigned in such a way that closest distance can be obtained from the distance matrix, which is optimum. When image element becomes equals to cluster number, it is replaced. PSNR value is much higher for segmented image compared to the data obtained from thresholding, which clearly speaks in favor of genetic algorithm. Cluster value and image matrix can be changed with the objective to reduce the fitness function.